Pacific Science (1973), Vol. 27, No.4, p. 399-405 Printed in Great Britain A Preliminary Report of the Biology of the Genus Charpentiera (Amaranthaceae) I S. H. SOHMER 2 ABSTRACT: The genus Charpentiera (Amaranthaceae), found in the Hawaiian and Austral archipelagoes, has a structurally gynodioecious but functionally dioecious breeding system. The sex ratio varies from taxon to taxon within the genus. The sex-determining mechanism is unknown. A high rate ofovule sterility is found in all the taxa, which i~ not predicted by the pollen sterility figures reported. Data concern­ ing seed size, pollen size, and seed germination potential are provided. Hybridization is demonstrated to occur between Charpentiera densiflora Sohmer and C. elliptica (Hilleb.) Heller. Reproductive isolation is reported for C. tomentosa var. tomentosa Sohmer and C. obovata Gaud. THE GENUS Charpentiera is indigenous to the often small and thinly scattered throughout the Hawaiian Islands. In 1934 it was also found mature ecosystems of the generic range. Popu­ on two of the Austral Islands approximately lations of several of the taxa, however, are 2,800 miles disjunct from Hawaii (Suessenguth sometimes numerous in certain areas such as 1936). The members of the genus are trees the heads of deep gulches in some of the highly ranging from about 8 to 40 feet high and are dissected volcanic ranges of the islands. During of diverse habit, with pendant, compound the course of the revision, the reproductive panicles of minute, anemophilous flowers. The biology of Charpentiera was investigated. fruit is indehiscent and uniovulate. A recent systematic revision of Charpentiera (Sohmer 1972) has recognized six major taxa in the SEX RATIO AND TENDENCY TO DIOECISM genus. Interpretation of certain morphological Charpentiera is gynodioecious. This fact was characteristics, habit, geographical distribu­ first realized by Skottsberg (1926) and later tion, and behavior as noted in the field in­ confirmed by Carlquist (1966). There are usu­ dicate that these taxa are best recognized at ally perfect-flowered individuals and pistillate the rank of species, three of which have been individuals in each population. That the genus described previously. Table 1 summarizes is in evolutionary transition toward a com­ several of the important systematic differences pletely dioecious reproductive system is indi­ among the taxa. cated by the frequent occurrence of undevel­ The populations of most of the taxa are oped ovules in the apparently perfect flowers and the fact that these flowers often abscise I This work is a revised portion of a dissertation pre­ sented in partial fulfillment of the Ph.D. degree in the after the pollen is shed. Therefore, functionally, Department of Botany, University of Hawaii. Work the members of the genus are often dioecious. with the genus was made possible by a University of Table 2 illustrates some of the variations in the Wisconsin Teachers Improvement Assignment in 1969­ sex ratios observed by me in the field. In many 1970. Fieldwork in the Hawaiian Islands was made possible partially by grants from the Hawaiian Botanical populations, all mature trees found were scored. Gardens Society Foundation and the McIntire-Stennis The low sample sizes reflect the small numbers project 674. Fieldwork with the genus on the Austral of individuals in many populations of Charpen­ Islands was made possible by National Science Foun­ tiera. The mechanism for determination of dation grant GB-28359. Manuscript received 18 January sexual types is not known. The ratio obtained 1973. 2 Department of Biology, University of Wisconsin at from the sum of the results for all populations La Crosse, La Crosse, Wisconsin 54601. studied in the genus yielded a chi-square con­ 26 399 H PS 27 400 PACIFIC SCIENCE, Volume 27, October 1973 TABLE 1 SELECTED CHARACTERS OF TAXONOMIC IMPORTANCE IN Charpentiera RANGE AND SAMPLE SIZE STEM FLORAL STEM WIDTH LEAF BLADE INTERNODAL INTERNODAL BEHIND LENGTH DISTANCE DISTANCE APEX TAXON HABIT LEAF SHAPE (cm) (mm) (mm) (mm) C. australis Tree to 35 elliptic to 10-22 (6) 0-3 (6) 0-2 (4) 3-5 (6) feet subobovate C. densiflora Robust tree to elliptic to 13-40 (27) 0-2 (13) 0-2 (15) 4-9 (12) 40 feet subobovate C. elliptica Slender tree to elliptic to 5-21 (35) 1-14 (27) 0.3-12 (28) 1.5-2.75 (31) 20 feet lanceolate C. obovata Short; shrubby obovate to 3-12.5 (115) 2-16 (83) 1.1-9 (81) 0.6-1.8 (92) tree to 12 feet elliptic C. tomentosa var. Robust tree obovate 9-23.5 (28) 0-5 (18) 0.8-5 (22) 2.6 (25) tomentosa to 40 feet C. tomentosa var. Slender tree obovate to 8.3-30 (79) 0-8 (29) 1-9 (36) 1.75-5 (41) maakuaensis to 20 feet subelliptic C. ovata var. Slender tree ovate 5.9-27 (59) 0-9 (22) 1-5 (14) 1.25-5 (41) ovata to 20 feet C. ovata var. Shrubby tree ovate to 5-18 (49) 1.6-14 (24) 1.5-5 (17) 0.5-2.5 (30) niuensis to 15 feet rotund TABLE 2 SEX RATIO IN Charpentiera RATIO PERFECT TO PISTILLATE TREES TAXON POPULATION (SAMPLE SIZE) C. densiflora Hoolulu Valley, Kauai 2:3(10) C. elliptica Nualolo Ridge, Kokee, Kauai 2:3(10) C. obovata Gulch below Puu Pane, Waianae 1: 2.75 (30) Mountains, Oahu C. obovata Kaluaa Gulch, Oahu 2:3 (10) C. tomentosa var. tomentosa Kaluaa Gulch, Oahu 1:10(11) C. tomentosa var. maakuaensis Maakua Gulch, Oahu 1: 3.2 (42) C. tomentosa var. maakllaensis Koloa Gulch, Oahu 1: 1.875 (23) C. ovata var. ovata Manoa Cliffs, Oahu 1: 3 (8) C. ovata var. niuensis Pia Gulch, Oahu 1: 1 (12) C. ovata var. nillensis Kului Gulch, Oahu only pistillate (10) ------------_.------ NOTE: The individuals ofC. australis available for study were sterile or immature. The data are from the author's field collections. sistent with the hypothesis of a 3: 1 pistillate to observed was in a relatively accessible gulch perfect ratio (P > 0.05). carved by the Maakua Stream near Hauula. To ascertain flowering behavior, I observed Ten trees were randomly selected and marked. individuals in a population on the island of At the appropriate time, about February to Oahu for a period of 1 year. The population March, as the new flowering season began, A Preliminary Report of the Genus Charpentiera-SoHMER 401 TABLE 3 POLLEN IRREGULARITIES IN Charpentiera PERCENTAGE TOTAL NO. OF PERCENTAGE WITH PERCENTAGE INDIVIDUALS WITH ABORTED PERCENTAGE WITH POPULATION SAMPLED* MICROPOLLEN POLLEN UNSTAINED IRREGULARITIES C. densiflora 4 0.6 0.6 Hoolulu Valley, Kauai C. elliptica 5 2.6 8.6 11.2 Nualolo Trail, Kauai C. obovata 7 2.9 2.9 Gulch below Puu Pane, Oahu C. obovata 9 0.3 0.8 1.1 Kaluaa Gulch, Oahu C. tomentosa var. maakllaensis 26 2.2 3.3 1.28 6.78 Maakua Gulch, Oahu C. tomentosa var. maakllaensis 7 0.1 0.8 0.9 Koloa Gulch, Oahu C. ovata var. ovata 7.6 15.2 22.8 Kipahulu Ridge, Maui C. ovata var. nillensis 6 1.5 0.1 1.6 Pia Gulch, Oahu * One hundred pollen grains were randomly selected and scored from each of five flowers from each individual sampled. pairs of young inflorescences were selected on three perfect-flowered individuals observed each tree. One member of each pair was en­ during the study, one possessed an inflorescence closed in a Duraweld 12-by-9-inch pollinating that showed mature seed after being bagged. bag and the other left as a control. The mesh In this particular individual, less than 5 percent on the Duraweld bags is less than the diameter of the approximately 400 flowers on the in­ of the pollen grains in Charpentiera. However, florescence possessed mature seed; most of the air and light may pass freely, the latter through other flowers had abscised. These seeds did not a plastic window. The bagged inflorescences have expanded contents. None of the trees that were observed every 2 weeks for 4 months. were marked demonstrated changes in the Thereafter, monthly visits to the site were sexual phenotype during the year of observa­ made. As flowering progressed, it was ascer­ tion. More experimentation of this type is tained that the sample comprised three perfect necessary before far-reaching conclusions can and seven pistillate trees. be drawn. However, at this point, we do have At the conclusion of the observations, no an indication that agamospermy does not occur. mature or developing seeds were found in the bagged inflorescences of pistillate trees, al­ though a few ovules had enlarged. The un­ blj,gged controls on these trees had abundant POLLEN STERILITY mature seed. Maturity was indicated by the The pollen studies are based upon my col­ development of a black testa. Undeveloped lections. Results obtained from other collec­ ovules retain their white color; and immature, tions are not given here. Perfect flowers at or developing ovules are reddish brown. The near anthesis were hydrated in boiling water perfect-flowered individuals either had abscised for a few minutes, and the anthers were dis­ their flowers after the anthers had dehisced or sected on a slide in a drop of an aniline-blue possessed gynoecia that did not mature. Of the lactophenol solution according to the formula 26-2 402 PACIFIC SCIENCE, Volume 27, October 1973 TABLE 4 POLLEN DIAMETER IN Charpentiera NO. OF POLLEN DIAMETER (p,) INDIVIDUALS TAXON SAMPLED MEAN RANGE C. densiflora 4 18.7 15.5-21.3 C. elliplica 5 22.4 20.0-25.3 C. obovala 16 18.9 16.3-22.5 C. lomenlosa var. lomenlosa 2 20.1 17.5-22.5 C.